Ann came to wealth management from a data and AI company, where she worked in customer success alongside deeply technical teams. She knows the vocabulary, knows what good reporting looks like, and knows her firm's data cold. What she lacked was a way to act on it without waiting in someone else's queue: “I could sit down and write the whole recipe and give it to an engineer, and he'd be like, ‘I guess I'll get to your report at some point.' Or I can just do it. And now I can.” She had already done the hard version — working with Claude directly, hand-writing context documents describing her firm's Snowflake schema so the model would understand what lived where. It worked, but every question began by re-explaining the environment. With MCP that groundwork is simply present, and metrics, tables, charts and the styling on top of them get built against real data. It reads and builds; it cannot alter the data underneath.